Human H+ATPase a4 subunit mutations causing renal tubular acidosis reveal a role for interaction with phosphofructokinase-1.

The vacuolar-type ATPase (H+ATPase) is a ubiquitously expressed multisubunit pump whose regulation is poorly understood. Its membrane-integral a-subunit is involved in proton translocation and in humans has four forms, a1-a4. This study investigated two naturally occurring point mutations in a4's COOH terminus that cause recessive distal renal tubular acidosis (dRTA), R807Q and G820R. Both lie within a domain that binds the glycolytic enzyme phosphofructokinase-1 (PFK-1). We recreated these disease mutations in yeast to investigate effects on protein expression, H+ATPase assembly, targeting and activity, and performed in vitro PFK-1 binding and activity studies of mammalian proteins. Mammalian studies revealed complete loss of binding between the COOH terminus of a4 containing the G-to-R mutant and PFK-1, without affecting PFK-1's catalytic activity. In yeast expression studies, protein levels, H+ATPase assembly, and targeting of this mutant were all preserved. However, severe (78%) loss of proton transport but less decrease in ATPase activity (36%) were observed in mutant vacuoles, suggesting a requirement for the a-subunit/PFK-1 binding to couple these two functions. This role for PFK in H+ATPase function was supported by similar functional losses and uncoupling ratio between the two proton pump domains observed in vacuoles from a PFK-null strain, which was also unable to grow at alkaline pH. In contrast, the R-to-Q mutation dramatically reduced a-subunit production, abolishing H+ATPase function completely. Thus in the context of dRTA, stability and function of the metabolon composed of H+ATPase and glycolytic components can be compromised by either loss of required PFK-1 binding (G820R) or loss of pump protein (R807Q).

transporting epithelia of the distal nephron contain a group of cells known as α-intercalated cells (α-ICs), which express a high density of H+ATPases on their apical plasma membranes. These cells are responsible for urine acidification in response to acidosis (14). Proper α-IC function is important not only for the maintenance of bodily acid-base homeostasis, but also for maintaining calcium solubility in urine and its stability in bone. Functional failure of α-ICs results in metabolic acidosis, which together with nephrocalcinosis and rickets are cardinal features of distal renal tubular acidosis (dRTA) (1).

In addition to their vital function in the kidney, H+ATPases pump protons across the plasma membranes of other cell types, such as osteoclasts and cells that line the epididymis and vas deferens, as well as being essential for acidification of many intracellular compartments in eukaryotic cells (5, 55).

H+ATPases from fungi, plants, and animals are structurally very similar, consisting of two major functional domains known as V1 and V0 (55). The V1 domain, comprised of eight different subunits (A-H), is responsible for ATP hydrolysis. The V0 domain contains at least five different subunits (a, c, c'', d, and e) and transports protons across the membrane (49, 55). This proton pumping requires both structural and functional coupling of the V1 and V0 domains. Assembly of the V1 and V0 subcomplexes into the holoenzyme is completed in the ER, the pump then being targeted to the required destination (15, 40). In yeast cells lacking a V1 domain, the V0 domain can still be assembled and targeted onto the vacuolar membrane. Conversely, if the V0 domain is absent, the V1 domain is assembled, but cannot be attached onto the membrane (reviewed in Ref. 55). In addition, free V1 and V0 domains lack MgATP hydrolytic and proton transport activity, respectively, indicating a requirement for coupling of V1 to V0 for normal pump function.

Yeast studies demonstrated that disruption of genes encoding any one subunit of the H+ATPase leads to a defect of vacuolar acidification, due to failure of either pump assembly or functional coupling between V1 and V0 domains (19, 41, 59). These deletion strains exhibit a conditional growth phenotype (Vma−) whereby they can only grow at acidic pH, and not at neutral or higher pH, nor in the presence of high levels of extracellular divalent cations, nor in media containing only nonfermentable sources of carbon (16, 41, 55).

The H+ATPase a-subunit, a component of the V0 domain, comprises a large NH2-terminal hydrophilic half, several transmembrane helices, and a small soluble COOH-terminal tail (20, 30, 47). Unlike all other subunits in yeast, two different a-subunit paralogs have been identified, Vph1p and Stv1p (38, 39). Vph1p resides in the vacuole, and yeast studies suggest that its COOH-terminal tail is critical not only for proton pumping, but also for stability of Vph1p, and assembly, targeting, and activity of the pump (24, 29, 44). In contrast, Stv1p resides in the Golgi. Both paralogs must be absent for yeast to exhibit the conditional (Vma−) phenotype (39).

In both mouse and humans, four different a-subunit paralogs (a1-a4) exist. Of these, a4 is expressed predominantly in the kidney, being located apically in α-ICs and subapically in the proximal tubule (45, 51–53). Defects in the genes encoding a4, the osteoclast-enriched a3, and endosomal a2 subunits are associated with recessively inherited dRTA (rdRTA), infantile malignant osteopetrosis, and cutis laxa, respectively (12, 27, 28, 52), underscoring the functional importance of this subunit in kidney, bone, and skin.

The gene encoding a4 was identified as the second causative gene for rdRTA (52), the first encoding B1, another renal H+ATPase subunit (22). A wide variety of homozygous mutations have been reported, most of which are predicted to disrupt the encoded protein (52, 56, 58). Only a few missense mutations have been described, two of which, R807Q and G820R, affect residues in the COOH terminus within a domain of the a-subunit that we previously reported is responsible for binding to PFK-1 (57).

To date, no laboratory has been able to achieve stable epithelial cell expression of the mammalian a-subunit; so to ascertain disease-causing mechanisms in these forms of rdRTA, we turned to yeast to investigate their effects on protein expression, H+ATPase assembly/targeting, and enzyme activity, together with the relationship between the a-subunit/PFK-1 interaction and H+ATPase function.

MATERIALS AND METHODS

Strain and plasmid construction.

KEBY9, a stv1Δ/vph1Δ double deletion strain of S. cerevisiae (23), was used for transformation of plasmid constructs. The coding regions of VPH1 or STV1 and their 5′ and 3′ UTRs amplified in-house by PCR were separately cloned into the low-copy (CEN) plasmid pRS316 to create constructs pKGB1 and pKEB4, respectively. Site-directed mutagenesis using the QuikChange Mutagenesis Kit (Stratagene) was performed using pKGB1 as template; base substitutions were separately introduced into codon 799 (CGT→CAA) or codon 812 (GGT→CGT), resulting in the single amino acid changes R→Q (pKGB1-R799Q) or G→R (pKGB1-G812R), respectively. All constructs were sequence-verified before use.

Yeast complementation assay.

KEBY9 cells were transformed with either pKGB1-R799Q, pKGB1-G812R, pKGB1, or empty vector as control, followed by selection on SD-U medium. Colonies were resuspended in 30 μl of H2O, which was then serially diluted and 5 μl of each dilution was pipetted onto YPD buffered to either pH 5, 7.5, or 7 with 100 mM CaCl2. Cell growth was scored following incubation at 25, 30, or 37°C for 3 days. Each assay was performed twice.

For experiments using the PFK-null yeast strain (Δpfk1:: LEU2.Δpfk2::URA2, gift of E. Boles, Frankfurt), 2% ethanol + 0.2% glucose replaced 2% glucose in the media, to provide a suitable carbon source while avoiding spontaneous H+ATPase disassembly (21).

Whole cell lysates and isolation of vacuolar membrane vesicles.

Yeast cells were grown to OD600 of 0.8 in SD-U medium at 30°C. Whole cell lysates were prepared using YeastBuster Protein Extraction Reagent (Novagen).

Vacuolar membrane vesicles were isolated essentially as described (48) with minor modifications: after growth in SD-U at 30°C to OD600 of 1–2, cells resuspended in 50 mM Tris·HCl, pH 9.5, containing 10 mM DTT were incubated at 30°C for 30 min. Cells were spheroplasted at 30°C for 90 min in 1.2 M sorbitol, 50 mM K3PO4 (pH 7.4), 5 mM MgCl2, 1% glucose, and 3.6 mg/ml zymolyase 100T. Spheroplasts were lyzed and vacuoles were purified by flotation on a discontinuous Ficoll 400 gradient. Purified vacuoles were resuspended in TE, pH 7.4, containing 1 mM PMSF and 1 μg/ml leupeptin.

Whole cell lysate and vacuole samples were analyzed by Western blot using mouse monoclonal antibodies 10D7 (α-Vph1p) and 8B1 (α-Vma1p, the yeast V1 domain's A subunit; Molecular Probes). Data were quantified densitometrically using ChemiImager V5.5 software.

Immunofluorescence microscopy.

Spheroplasting, fixation, and mounting of yeast cells were performed as described (48). Following blocking for 1 h with 1% normal goat serum in PBS/BSA (5 mg/ml), cells were incubated first with rabbit polyclonal α-Vph1p antibody (18) (gift of T. Stevens to K. Bowers) at room temperature (RT) for 2.5 h, then biotin-conjugated goat α-rabbit (Jackson Immuno Research) secondary for 1 h, followed by 1 h with streptavidin-conjugated fluorescein, DTAF (Jackson ImmunoResearch). Slides were mounted in Vectashield Mounting Medium with DAPI (Vector Labs). All images were collected with a Zeiss Axiovert 200M microscope.

H+ATPase functional assays.

Bafilomycin-sensitive ATPase activity of yeast vacuoles was measured as described (13). Three micrograms of vacuoles were added into 1 ml of assay mixture [50 mM KCl, 50 mM Tris, pH 8.0, 2 mM MgCl2, 0.2 mM NADH, 1 mM phosphoenolpyruvate, 2 mM MgATP (pH 7.4), 10–11 U/ml lactate dehydrogenase, and 12–13 U/ml of pyruvate kinase (Sigma)] at 25°C. The corresponding decrease in absorbance of NADH at 340 nm was monitored for 4 min, with and without the addition of 0.3 μM H+ATPase inhibitor bafilomycin A using a UV-1601 spectrophotometer.

ATP-dependent proton transport activity of yeast vacuole samples was measured using fluorescence quenching of acridine orange (Sigma) by 3–4 μg vacuoles in buffer [20 mM Mops/Tris, pH 7, 150 mM KCl, 4 μM acridine orange, 1 mM Mg/ATP, with and without bafilomycin A (0.3 μM)] and corrected for mass, as described (43).

These H+ATPase functional properties were measured in triplicate in each of two separate vacuole preparations. For each type of assay, comparison of wild-type (WT) with mutant activities was made by taking WT levels as 100%.

Peptide synthesis.

Peptides corresponding to the coding sequences of the last 45 residues of WT human a4 (a4C-WT; Fig. 1) or a4C containing G820R (a4C-GR) with acetylated NH2 termini were synthesized and HPLC-purified by CovalAb UK (Cambridge, UK).

Fig. 1.

Confirmation of yeast H+ATPase/Pfk interaction and comparison of a-subunit COOH-terminal sequences. A: Co-IP of H+ATPase and Pfk from vacuoles of KEBY9 (a-subunit double deletion) yeast transformed with Vph1p. IP: (+) or (−) indicates inclusion or omission of α-Vma2p; IB: α-Pfk. A single band corresponding to yeast Pfk is seen in the (+) lane only. B: amino acid sequences of the COOH termini from human and yeast orthologs and paralogs are aligned. Residues at the start and end of this domain are numbered. The highly conserved first 17 amino acids are highlighted in gray. Conserved Arg (807 in a4 or 799 in Vph1p) and Gly (820 in a4 or 812 in Vph1p) residues are boxed.

Expression and purification of PFK-1.

PCR-amplified and sequence-verified cDNA encoding full-length rabbit muscle-type PFK-1 was cloned into pET5a plasmid and expressed in the PFK-deficient Escherichia coli strain DF1020 (7), which was modified by lysogenization with lambda (DE3) prophage (Novagen) at 30°C for 6 h. Cells were lysed by sonication and PFK-1 was purified first through a Cibacron blue 3GA-agarose column (Sigma) and then by ammonium sulfate precipitation and finally by an HiTrap Q HP anion exchange column (Amersham Biosciences). Purity, specificity, and activity of the recombinant PFK-1 were analyzed by SDS-PAGE, Western blot using a goat polyclonal anti-PFK-1 antibody (Chemicon Europe), and enzyme activity assays, respectively (Supplementary Fig. 1), before use (the online version of this article contains supplemental data).

Surface plasmon resonance analysis.

Binding affinity analysis was performed on a BIAcore 2000 biosensor system (Pharmacia Biosensor AB) using surface plasmon resonance (SPR) measurements. A carboxymethylated CM5 sensor chip was activated with a 1:1 mixture of 0.4 M N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide and 0.1 M N-hydroxysuccinimide. Recombinant rabbit muscle-type PFK-1 (60 μg/ml in 10 mM NaOAc, pH 5) was then immobilized on the sensor chip by amine-coupling according to the manufacturer's instructions. Unreacted sites were blocked with 1 M ethanolamine/HCl (pH 8). Control flow cells were activated and blocked in the absence of PFK-1. Flow cells were routinely equilibrated with running buffer (PBS, 0.005% surfactant P20). Each synthetic a4C peptide was diluted in this buffer and allowed to interact with the sensor surface by a 180-s injection. Five to one hundred seventy five micromolar each a4C peptide was injected, each at a flow rate of 10 μl/min at 25°C. Data from duplicate assays were modelled for binding equilibria using Origin (OriginLab software).

Circular dichroism spectroscopy.

Secondary structures of a4C peptides were analyzed with far-UV circular dichroism (CD) spectroscopy. Spectra were measured using a Jasco J-810 spectropolarimeter at 25°C. Each peptide was dissolved in 20 mM phosphate buffer (pH 7) to a final concentration of 25 or 50 μM before being transferred into a measuring cuvette (cell length 0.2 mm). Wavelength scans were collected in 0.2-nm decrements from 260 to 180 nm with a 1-nm bandwidth and a 2-s time constant at 50 nm/min. Ten scanning spectra were averaged for each sample. Estimates of secondary structural components were obtained using CDNN (version 2.1) (3).

PFK-1 activity assay.

PFK-1 activity measurements were conducted by coupling the reaction catalyzed by PFK-1 to oxidation of NADH and monitoring the corresponding decrease in absorbance at 340 nm at 25°C (10, 31). Reaction mixtures contained 50 mM Tris·HCl, pH 8, 5 mM MgCl2, 1 mM DTT, 0.2 mM NADH, 40 μg/ml aldolase, 2 μl/ml triosephosphate isomerase + glycerophosphate dehydrogenase readymade mixture (Roche), 10 μg/ml recombinant rabbit muscle-type PFK-1, 1 mM ATP, and 0.0275 to 0.305 mM fructose-6-phosphate (F6P). To investigate the potential effects of a4C on PFK-1 activity, each a4C peptide was preincubated with PFK-1 for 2 h at RT before inclusion in the reaction mixture. Each peptide was assayed on three different occasions. Kinetic data were analyzed using Origin.

Coimmunoprecipitation.

Thirty-five micrograms of the vacuoles isolated as above from KEBY9 cells transformed with VPH1-containing plasmid were resuspended in binding buffer (TBS, 1% NP40, 10% glycerol) and incubated overnight at 4°C with 20 μl of 13D-11 monoclonal antibody (Molecular Probes) against Vma2p, the B subunit of the V1 domain. Antibody was omitted in parallel control experiments. After 1-h incubation with α-mouse IgG-agarose beads (Sigma), the immunoprecipitates were washed three times with buffer A (TBS, 1% NP40, 5 mM NaN3); three times with buffer A plus 500 mM NaCl; and finally three times with buffer A. Bound proteins were eluted from beads in SDS sample buffer (0.18 M Tris·HCl, pH 6.8, 5.14% SDS, 18% glycerol, 0.3 M DTT, 0.006% bromophenol blue) at 95°C for 5 min and supernatants were subjected to SDS-PAGE. Western blot was performed with an antibody against yeast Pfk (gift of J. Heinisch, Osnabrüeck) (17).

RESULTS

The ATP6V0A4 mutations investigated in this study were previously reported as homozygous alterations in two different patients diagnosed with rdRTA: G820R in patient 5-1 (52) and R807Q in patient 70-1 (56) (Supplementary Table 1). They presented at 11 and 2 mo of age, respectively, with normal anion gap metabolic acidosis and inappropriately alkaline urine, fulfilling diagnostic criteria. Following the original report, audiometry at age 9 in patient 5-1 revealed no evidence of hearing loss, whereas by this age, severe loss was present in patient 70-1 who is now profoundly deaf.

We wished to express human a4 in the vph1Δ/stv1Δ double deletion yeast strain KEBY9, in which expression of either Vph1p or Stv1p restores growth (39). For initial validation of a yeast model, we coimmunoprecipitated the vacuolar H+ATPase and Pfk from KEBY9 reconstituted with Vph1p, to confirm that the two are associated in yeast as well as in mammalian tissue (Fig. 1), observing a band of the correct size for yeast Pfk in the + lane only. Unfortunately, WT human a4 constructs in two different vectors could not complement KEBY9 growth under several conditions (Supplementary Fig. 2). Instead, we took advantage of the high homology of this region across species (Fig. 1) and recreated the mutations R799Q and G812R in conserved residues of Vph1p for these studies. Compared with WT Vph1p transformants, yeast cells carrying the R799Q mutant could not rescue the deletion phenotype (Fig. 2). However, cells containing the G812R mutant protein showed no obvious difference in growth compared with WT under all conditions tested, indicating complementation. Using whole cell lysates of each sample, Western blot using α-Vph1p antibody demonstrated that the level of G812R mutant protein was comparable to WT Vph1p (Fig. 2). In contrast, an 86% reduction of R799Q mutant protein was observed, in keeping with the complementation results.

Fig. 2.

Effect of R799Q and G812R mutations on yeast growth, Vph1p expression, H+ATPase assembly, and targeting. A: KEBY9 yeast cells were transformed with relevant constructs containing one of either VPH1-R799Q (R>Q), VPH1-G812R (G>R), VPH1 (Vph1p), or empty vector. Images are of 30°C cultures. Functional complementation was observed from the G812R mutant at a level comparable to wild-type (WT) Vph1p, but not by the R799Q mutant. Yeast whole cell lysates (B) or vacuolar membrane vesicles (C) isolated from cells expressing WT or mutant forms of Vph1p, or control KEBY9 cells (Δ/Δ), were analyzed by Western blot using antibodies against Vph1p and/or Vma1p, densitometrically quantified, and expressed relative to WT (100%) ± SE. B: whole cell lysate showed expression of both mutants, with G812R at WT levels and R799Q severely reduced. C: relative amounts of each mutant present on the vacuolar membranes mirrored that seen for whole cells. Both V0 and V1 subunits (Vph1p and Vma1p) were present in G812R and R799Q mutant vacuoles, indicating that each was capable of assembly and targeting (top).

To test the effect of the mutations on assembly and targeting of the H+ATPase, isolated vacuoles were first analyzed by Western blot using α-Vph1p and α-Vma1p antibodies, to identify a V0 and a V1 component, respectively. As shown in Fig. 2, G812R mutant vacuoles showed levels of both Vph1p and Vma1p comparable to WT cells, suggesting that this mutation does not impair assembly or vacuolar targeting of the intact H+ATPase, for which both V1 and V0 domains must be present. The reduced level of Vph1p observed in R799Q mutant vacuoles was similar to those found in the whole cell preparation but was still accompanied by Vma1p, suggesting that the small amount of mutant R799Q protein produced is capable of being assembled and targeted onto vacuolar membranes.

Concomitant immunocytochemistry showed that similar to WT Vph1p (Fig. 3, A–C), the G812R mutant was expressed and localized at the vacuolar membrane (E and F) in almost every cell (D), further confirming normal expression and targeting of these mutant pumps. Predictably, the majority of R799Q mutant cells showed almost no Vph1p staining (Fig. 3). However, a small population of cells did show some vacuolar membrane localization of mutant Vph1p (Fig. 3, H and I), in good agreement with data in Fig. 2.

Fig. 3.

Indirect immunofluorescence examination of Vph1p expression and targeting. KEBY9 cells transformed with plasmid carrying either WT or mutant VPH1 were fixed, spheroplasted, and stained with polyclonal α-Vph1p antibody (green). To observe vacuoles, differential interference contrast (DIC) images were collected (right, C, F, I, L). DAPI-labeled nuclei are blue. Left: expression of R799Q (G) is much lower than WT (A) and G812R (D), consistent with Western blot data in Fig 2. At higher magnification (middle), images (B, E, H) confirm that both mutant proteins can reach the vacuolar membrane. Control KEBY9 cells (Δ/Δ) show no Vph1p (J, K). Scale bars = 3 μm.

We next investigated both ATPase and proton translocation activity of R799Q or G812R mutant pumps. As expected from the low protein levels, the R799Q mutation led to almost complete loss of both bafilomycin-sensitive ATPase and proton transport activity, indicating a lethal effect of this mutation on H+ATPase function (Fig. 4). In marked contrast, the G812R mutant lost 36 ± 5% of WT levels of bafilomycin-sensitive ATPase activity, but a much greater fraction of proton transport (78 ± 4%). The significant difference between these (P = 0.023 by t-test) is indicative that this mutation leads to loss of the required functional coupling between proton transport (V0) and ATP hydrolysis (V1), but preserves enough proton translocation to permit yeast cell growth at alkaline pH. Interestingly, the severe loss of proton translocation observed for this mutant is similar to that reported in an earlier study of a3 and a4 subunit mutants (44) but there, ATPase activity was also severely impaired. The authors used a different yeast strain, in which Vph1p but not Stv1p was deleted, and this difference is one feature that may account for differing results.

Fig. 4.

Effects of R799Q and G812R mutations on H+ATPase activity. Bafilomycin-sensitive ATPase activity and rate of ATP-dependent proton transport in vacuolar membrane vesicles isolated from cells expressing WT or mutant forms of Vph1p are shown as percentages of WT levels ± SE. The specific activity of bafilomycin-sensitive ATPase activity of vacuoles containing WT-Vph1p was 1.02 ± 0.09 μmol ATP hydrolyzed·min−1·mg protein−1. Approximately 70% of the ATPase activity of WT vacuoles was bafilomycin sensitive. Experiments were performed in triplicate in each of 2 preparations per strain. Supplementary Fig. 4 shows representative proton translocation traces.

In the face of preserved expression, assembly, and targeting of this mutant H+ATPase, we hypothesized that the functional uncoupling and severe reduction in pump activity engendered by the G to R substitution might result from disruption of the a4/PFK-1 interaction. To test this, we expressed, purified, and confirmed the enzymatic activity of full-length rabbit muscle-type PFK-1 (Supplementary Fig. 1), synthesized WT and mutant peptides corresponding to the last 45 amino acids of human a4 (a4C-WT and a4C-GR), and then performed SPR analysis using the BIAcore system.

BIAcore measurements are expressed in resonance units (RU) proportional to the concentration of immobilized protein. Figure 5, A and C, displays sensorgrams of a4C-WT (top) or a4C-GR mutant peptide (bottom) binding to PFK-1 and Fig. 5, B and D, shows the corresponding binding curves. The Kd value for a4C-WT/PFK-1 interaction was 29.4 ± 6.2 μM. A similar value was obtained using rabbit liver-type PFK-1 (Sigma) in place of the recombinant muscle-type PFK-1 (data not shown). However, the Kd value for a4C-GR/PFK-1 interaction could not be generated due to extremely low affinity, indicating severe disruption of binding.

Fig. 5.

Surface plasmon resonance (SPR) assay of a4C/PFK-1 interaction. Synthetic human a4C peptides in the range 5–175 μM were analyzed for binding to immobilized rabbit PFK-1. Representative sensorgrams are displayed at left (A, C) and show amount of bound a4C-WT or a4C-GR (RU) over time. B and D: binding curves reveal saturable kinetics for the a4C-WT/PFK-1 interaction (Kd = 29.4 ± 6.2 μM), whereas a4C-GR and PFK-1 interacted too weakly to permit calculation.

We then asked whether this disruption might be explained by secondary structural alteration in a4C-GR. The CD spectra of both WT and mutant a4C peptides exhibit the two large negative bands ∼208 and 222 nm suggestive of mainly α-helical structure in solution. Figure 6 shows data for 50 μM peptides; data for 25 μM analysis were not different. A rightward shift was evident in the a4C-GR mutant spectrum, and spectral analysis using the program CDNN suggested an increase in random coils of almost 50%, with a concomitantly decreased α-helical content in the mutant (Table 1), indicating some loss of regular structure.

Fig. 6.

Far-UV circular dichroism (CD) spectra of a4C-WT and a4C-GR. Each is the average of 10 spectra at 50 μM. Spectral shifts were the same for 25 μM peptides. The G820R spectrum is right-shifted compared with WT, suggesting a change in secondary structure. See Table 1 for structural modelling.

Table 1.

Percentage of secondary structure elements present in a4 COOH-terminal 45 amino acids, estimated from 10 CD spectra using CDNN software

Peptides	a4C-WT	a4C-GR
α-Helix	41%	33%
β-Sheet	12%	16%
β-Turn	16%	16%
Random Coil	26%	38%

WT, wild-type.

To exclude a direct effect of the a4C-GR peptide on enzyme activity of PFK-1, we measured the comparative activity of PFK-1 in phosphorylating F6P with and without inclusion of WT or mutant a4C. a4C-WT peptides ranging between 3 and 50 μM were initially preincubated with PFK-1 and then incorporated into a reaction mixture containing an excess of F6P; similar activity levels were found (Supplementary Fig. 3). Thereafter, experiments employing 6 μM a4C with varying amounts of F6P generated saturation curves and kinetic parameters that best fitted the first-order Michaelis-Menten equation (Fig. 7 and Table 2). Specific activity and Km values for a4C peptide-bound PFK-1 were not significantly different from that of free PFK-1 alone (P > 0.45), confirming that neither WT nor mutant a4C peptide affected PFK-1's catalytic ability.

Fig. 7.

PFK-1 activity assays. Enzyme activity was assayed with and without a4C-WT or a4C-GR. These peptides did not affect activity. Representative curves shown are means of triplicate replicates in 1 of 3 experiments. Curves fitted first-order kinetics (Origin software). See Table 2 for kinetic parameters.

Table 2.

Steady-state kinetic parameters for PFK-1 ± WT or mutant a4C peptides

Proteins	Km(F6P, mM)	Vmax, mol/min
PFK-1	0.16±0.03	2.61±0.19
PFK-1 + a4C-WT	0.13±0.02	2.42±0.15
PFK-1 + a4C-GR	0.11±0.02	2.28±0.13

Data are means ± SE of 3 assays. No significant differences were found.

Finally, to confirm the role of PFK in proton pump regulation, we employed a Saccharomyces strain lacking this enzyme. Yeast Pfk is a hetero-octamer containing two subunits, α and β, encoded by PFK-1 and -2, respectively. The Pfk1/Pfk2 null strain requires a nonfermentable carbon source such as ethanol for growth. In the presence of both ethanol and a small amount of glucose (to prevent H+ATPase disassembly), Fig. 8 demonstrates that like yeast lacking both proton pump a-subunits, the absence of Pfk imposes a pH-conditional growth phenotype whereby growth cannot proceed at raised pH, indicating defective vacuolar function. Second (Fig. 8), vacuoles isolated from this strain behaved similarly to those from the G812R mutant strain, demonstrating similar loss of bafilomycin-sensitive ATPase activity (39 ± 1%) compared with WT S. cerevisiae, with again a much more severe accompanying proton translocation loss (83 ± 2%), confirming Pfk's role as a cofactor for proton pump function and necessary mediator of functional coupling between V1 and V0 domains.

Fig. 8.

H+ATPase functional disruption in PFK-null yeast. Similar to the strain lacking H+ATPase a-subunits, acid-dependent growth is exhibited (A), and bafilomycin-sensitive ATPase activity and proton translocation rates are impaired in PFK-null yeast (B), confirming that Pfk is essential for normal proton pump function.

DISCUSSION

H+ATPases form complexes with glycolytic enzymes; we previously identified an interaction between the COOH-terminal tail of the human a-subunit and PFK-1 (57), and the E subunit has been found to interact with aldolase (35). Since the integrity of proton pump function is dependent on the presence of glucose (21), these observations provided a potential regulatory mechanism for H+ATPase in higher eukaryotes, and highlight the functional importance of the a-subunit's COOH terminus. By seeking insight into mechanisms of disease caused by the two known disease-causing mutations (R807Q and G820R) in the COOH terminus of a4, we here established an important role for PFK-1 in the proton pump complex.

Both R807 and G820 are highly conserved across species, and both lie within a region that is similarly highly conserved. While the R799Q mutation in the yeast model resulted in loss of protein and therefore function, which is commonly found in recessive disease, the mechanism underlying rdRTA associated with the G820R substitution is the more complex, affecting neither stability of the encoded protein nor assembly and targeting efficiency of the H+ATPase. Instead, the observed significant reductions in both ATP hydrolysis and more importantly, proton translocation, were associated with a dramatically lowered binding affinity between the mutant a-subunit tail and PFK-1, such that the interaction was too weak to generate kinetic data. This disruption of binding could have resulted from the change of the residue's charge and/or from the change in secondary structure of the a4C region as suggested by CD spectroscopy analysis.

Coupling of proton transport and ATP hydrolysis (subserved by V0 and V1 domains, respectively) is essential for normal overall function of H+ATPases. Intrinsic subunits from both domains (namely A, C, H, a, and d) have been reported to play a role in this functional coupling (4, 6, 42, 50). Although changes in coupling efficiency are recognized as an important mechanism for regulating H+ATPase activity, being affected by nonprotein factors such as ATP, trypsin, or NaN3, the normal intracellular signals controlling coupling remain unclear (11). Our finding of a role for PFK, a known pump-associated protein, provides new insight into the regulation of H+ATPase activity.

Having confirmed that in yeast, as in humans, the H+ATPase and PFK are associated, we have gone on to demonstrate that the Pfk-null yeast vacuoles behave similarly to the G812R mutant strain, with much more severe impairment of proton transport than of ATPase activity, showing the necessity of this glycolytic enzyme for functional coupling of the different domains of the pump and supporting the concept that loss of function and uncoupling caused by the G812R substitution are directly associated with loss of binding of PFK-1 to a4C-GR. It is notable that whereas sufficient proton pump activity was present to permit rescue of the conditional growth phenotype in G812R mutant yeast, complete absence of Pfk did not permit growth under these conditions. This is consistent with previous studies demonstrating that a mutant strain with as little as 20% preservation of WT pump activity is sufficient to rescue the growth phenotype, whereas less than 20% of WT activity would display a Vma− phenotype (33, 37). In addition, complete absence of the glycolytic component may exert a greater overall metabolic derangement, for example an alternative carbon source to glucose must be provided to permit any growth at all.

Evidence supporting the involvement of glycolysis in regulation of H+ATPase function is not limited to our findings. Aldolase (the next enzyme down the glycolytic pathway) interacts with the H+ATPase via the E-subunit, but the mechanism is different, affecting assembly of V1 and V0 domains rather than their functional coupling (34, 36). Earlier assessment of proton transport in the isolated turtle urinary bladder showed that it could be driven by the energy from both aerobic and anaerobic glycolysis (2, 54). Also, investigation of the effect of glucose on the reversible assembly of the V1 and V0 domains of the pump complex in yeast suggested not only coupling between H+ATPase activity and glycolysis, but also that further glucose metabolism is required, since accumulation of glucose-6-phosphate was insufficient to maintain or induce pump assembly (46). Thus there are several means by which PFK-1 may function as an indirect regulator of the proton pump.

During glycolysis, PFK-1 catalyzes the phosphorylation of F6P by MgATP to form fructose 1,6-bisphosphate and MgADP. Its activity is allosterically controlled by several activators and inhibitors (8). The active sites identified for substrate binding and allosteric regulation involve residues mainly located in the NH2-terminal half of PFK-1 (25, 26, 32, 60), whereas the binding region for the a-subunit that we identified is at the COOH-terminal end of PFK-1 (57). By comparison with the bacterial crystal structure, we could assume a priori that binding of a4 to PFK-1 would not affect its enzyme catalytic activities, and indeed, activity was not adversely affected by a4C. Thus we could exclude the possibility that abolishing the a4C/PFK-1 interaction exerts a direct effect on PFK-1 activity leading to secondary loss of H+ATPase function.

Extrapolating between species does require some caution, and while we would have preferred to perform these studies in an epithelial system, no group has successfully established stable expression of mammalian a-subunits. In our studies, we utilized human constructs wherever possible, for example in the SPR and PFK enzymatic assays. The vph1Δ/stv1Δ mutant yeast strain is, however, a useful tool for the study of the a-subunit (39), as it can be rescued by restoration of either gene back into the double deletion strain, where it reassembles with the other subunits to reform functional H+ATPases. Thus transformation with WT or mutants of Vph1p alone, which normally resides in the vacuole, provides an effective cellular model.

Our results for R807Q are in good agreement with the data obtained from a yeast mutagenesis study that employed different substitutions of this residue (24), in which changing the Arg to Leu, Ala, or Lys resulted in variable functional effects: dramatic destabilization of Vph1p, or a severe defect in assembly and activity of H+ATPase, or inactivated the pump without affecting assembly. In contrast, the earlier report of mutagenesis correlating to G820R in a yeast model noted above suggested a more severe decrement in ATPase activity than we observed, but similar compromise of proton translocation (44). Despite overall more severe functional compromise reported there, the mutant strains were still able to grow at alkaline pH in both studies. Possible reasons for the difference include methodological variations (for example, temperature, inhibitor, or buffer components) as well as the strain difference mentioned earlier. We ensured that the linear phase of enzyme activity was assessed at constant temperature, using reagents and equipment frequently in use and validated, and obtained similar and highly reproducible results in all assays.

We were interested to consider the effects on vacuolar ATPase function of replenishing the PFK-null strain with a variant lacking the COOH-terminal domain that interacts with a4C. However, it has already been reported that such constructs result in only poor levels of protein expression, which would make interpretation impossible (9). Importantly, it has been previously demonstrated that the pump spontaneously disassembles when no glucose is present (21); thus for our assays, we were careful to include a small amount of glucose in the media in addition to the ethanol required by this null strain, to avoid confounding results.

Considering the phenotypic consequences of the two mutations in the patients themselves, the patient carrying two copies of the R807Q mutation has been much more severely affected than the child who is homozygous for G820R, with a younger age at diagnosis, more severe acidosis, and significant deafness. This difference is reflected in the differing effects of the mutations, since the R799Q mutation did not generate sufficient protein for any significant functional activity, whereas the G812R mutant was able to do so. Interestingly, despite its extremely low levels, almost all of the R799Q mutant produced appeared at the vacuolar membrane, implying this mutation per se does not affect assembly and targeting of the H+ATPase, but rather protein stability with an overall ablative effect on the pump. Thus this Arg residue is critical for stable expression of the a-subunit and therefore its function.

Hearing impairment is highly variable among those carrying destructive recessive mutations in a4, but the basis of the apparent redundancy of a4 function in the inner ear in some patients is not clear. The child with predicted complete loss of a4 function due to R807Q is severely hearing impaired, whereas hearing is at present preserved in the patient with the G820R change, who is now 17. In addition, her alkali requirement has progressively diminished as she has reached teenage. A formal dietary assessment has not been carried out to determine whether the difference can be accounted for by diet or by the milder course of her disease. Previous mutagenesis studies of different subunits of yeast H+ATPase demonstrate that growth can be observed if a reassembled mutant H+ATPase enzyme has as little as 20% of WT activity (33, 37), and since the milder mutant retained about this level of function in vitro, we can conclude that this level of activity is sufficient to maintain inner ear fluid pH homeostasis, but not to acidify the urine adequately, at least during childhood. In addition, this particular patient may yet lose hearing in young adulthood, in common with others with reported ATP6V0A4 mutation (56).

In summary, these data demonstrate an important role for PFK-1 in normal proton pump function, that loss of a-subunit/PFK-1 interaction is likely to decrease the stability of the “metabolon” formed by various H+ATPase subunits (notably E and a) and glycolytic components (aldolase and PFK-1) (35, 57), and also that differing mechanisms for rdRTA underlie these point mutations in a4.

GRANTS

This study was funded by the Wellcome Trust. B. Javid was a Medical Research Council Training Fellow.